Developmental regulation of locomotive activity in Xenopus primordial germ cells


  • Kohei Terayama,

    1. Department of Life Science, Graduate School of Life Science, University of Hyogo, Akou-gun, Japan
    Search for more papers by this author
  • Kensuke Kataoka,

    1. Department of Life Science, Graduate School of Life Science, University of Hyogo, Akou-gun, Japan
    Current affiliation:
    1. Institute of Molecular Biotechnology of the Austrian Academy of Sciences, Vienna, Austria
    Search for more papers by this author
  • Keisuke Morichika,

    1. Department of Life Science, Graduate School of Life Science, University of Hyogo, Akou-gun, Japan
    Current affiliation:
    1. Department of Life Science, Faculty of Science, Rikkyo University, Toshima-ku, Tokyo, Japan
    Search for more papers by this author
  • Hidefumi Orii,

    1. Department of Life Science, Graduate School of Life Science, University of Hyogo, Akou-gun, Japan
    Search for more papers by this author
  • Kenji Watanabe,

    1. Department of Life Science, Graduate School of Life Science, University of Hyogo, Akou-gun, Japan
    Search for more papers by this author
  • Makoto Mochii

    Corresponding author
    • Department of Life Science, Graduate School of Life Science, University of Hyogo, Akou-gun, Japan
    Search for more papers by this author

Author to whom all correspondence should be addressed.



Primordial germ cells (PGCs) arise in the early embryo and migrate toward the future gonad through species-specific pathways. They are assumed to change their migration properties dependent on their own genetic program and/or environmental cues, though information concerning the developmental change in PGC motility is limited. First, we re-examined the distribution of PGCs in the endodermal region of Xenopus embryos at various stages by using an antibody against Xenopus Daz-like protein, and found four stages of migration, namely clustering, dispersing, directionally migrating and re-aggregating. Next, we isolated living PGCs at each stage and directly examined their morphology and locomotive activity in cell cultures. PGCs at the clustering stage were round in shape with small blebs and showed little motility. PGCs in both the dispersing and the directionally migrating stages alternated between the locomotive phase with an elongated morphology and the pausing phase with a rugged morphology. The locomotive activity of the elongated PGCs was accompanied by the persistent formation of a large bleb at the leading front. The duration of the locomotive phase was shortened gradually with the transition from the dispersing stage to the directionally migrating stage. At the re-aggregating stage, PGCs became round in shape and showed no motility. Thus, we directly showed that the locomotive activity of PGCs changes dynamically depending upon the migrating stage. We also showed that the locomotion and blebbing of the PGCs required F-actin, myosin II activity and RhoA/Rho-associated protein kinase (ROCK) signaling.


In many animals, primordial germ cells (PGCs) are specified at a distance from the future gonadal region at early embryonic stages and subsequently migrate toward the gonadal anlagen via a pathway specific for each species (Molyneaux & Wylie 2004). For example, mouse PGCs are specified in the proximal epiblast, migrate through the primitive streak, the parietal endoderm, the allantois, the hindgut and the mesentery, and finally arrive at the genital ridges (Bendel-Stenzel et al. 1998). Zebrafish PGCs emerge within the deep blastoderm, migrate anteriorly or laterally, align within the lateral mesoderm, and finally colonize the gonadal anlagen (Raz 2002). In Xenopus, PGCs arise as germ plasm-bearing cells in the vegetal region of the early embryo. At the tailbud stage, they begin to migrate from deep in the endoderm toward the most dorsal region (Kamimura et al. 1976, 1980; Nishiumi et al. 2005). Then they translocate to the mesentery and arrive at the gonadal ridge (Whitington & Dixon 1975).

As PGCs migrate long distances in embryos, they are assumed to have high motility. In fact, it has been reported that migrating PGCs showed active locomotive behavior in vivo, in slice cultures, and in vitro (Wylie & Heasman 1976; Heasman & Wylie 1978; Reichman-Fried et al. 2004; Blaser et al. 2005; Dudley et al. 2007). On the other hand, PGCs are also assumed to change dynamically their motility and/or migratory mechanism as they migrate within different types of tissue and proceed through multiple steps, including the initiation of migration, directional migration and termination at destination. To better understand the mechanism underlying PGC migration, it is necessary to define exactly the migration stages in vivo and evaluate the intrinsic motility of the PGC at each stage. For this purpose, the tailbud embryo of Xenopus laevis provides an ideal system. Injection of mRNAs flanked with the DEADSouth 3′ untranslated region (UTR) enabled us to follow PGCs in living embryos and to manipulate gene functions in a PGC-specific manner (Kataoka et al. 2006; Morichika et al. 2010; Takeuchi et al. 2010). Furthermore, after dissociation of embryos, labeled PGCs can be isolated and cultured easily (Morichika et al. 2010). In this study, we found the four successive stages of PGCs in the endodermal region of tailbud embryos, that is, clustering, dispersing, directionally migrating and re-aggregating. We then cultivated PGCs isolated from embryos at each stage and revealed that their morphologies and locomotive properties changed according to developmental stage. We also showed that the migrating PGCs developed and maintained a large bleb and that the locomotion and bleb's formation were dependent on F-actin, myosin activity and RhoA/ROCK signaling.

Materials and methods

DNA construction and mRNA synthesis

Venus-DS/pCS2 encoding the Venus fluorescent protein flanked by the DEADSouth 3′ UTR has been described previously (Kataoka et al. 2006). Full-length cDNAs of RhoA (GenBank ID: AF151015), Rac1 (GenBank ID: AF174644) and Cdc42 (GenBank ID: AF275252) were amplified from cDNA of Xenopus stage 32 embryos by polymerase chain reaction (PCR) using the primers gcgggatccatggcagccattcgtaagaag and gttgaattcttagatgagaaggcaacgtgc, attggatccatgcaggccattaaatgtgtg and tgggaattcttacaagagcagacattttct, and attggatccatgcagacaattaaatgtgta and gaggaattctcatagcagcatacacttgcg, respectively. The PCR products were cloned into the pCRII-TOPO vector (Invitrogen). Dominant negative forms of Rac17N, RhoA19N and Cdc4217N, were generated by site-directed mutagenesis from the corresponding full-length cDNAs according to a previous study (Habas et al. 2003). 5′ fragments and 3′ fragments for each mutated cDNA were amplified separately and ligated into a plasmid CS2 containing the DEADSouth 3′ UTR. mRNAs were synthesized with plasmids linearized by Not1 using a mMESSAGEmMACHINE kit (Ambion), and purified with the S-400 Spin Column (GE Healthcare) for microinjection.

Embryo preparation and microinjection

Xenopus laevis embryos were obtained by in vitro fertilization and de-jellied with 3% cysteine hydrochloride (pH 7.8) according to a standard method (Kataoka et al. 2006). Embryos were maintained in 0.1 × Marc's modified Ringer (MMR) at 14°C until the two-cell stage, and then cultured at 18°C. The developmental stages were determined according to Nieuwkoop & Faber (1994). mRNA synthesized in vitro was injected into the vegetal cortex of one-cell embryos as described previously (Kataoka et al. 2006).

Isolation and cultivation of PGCs

An embryonic region that contained most of the endoderm was excised using forceps from the Venus-DS-injected embryos at stage 18, 24, 28, 33/34, 37/38 or 41 and incubated in 70% Dulbecco's phosphate-buffered saline without Ca2+ and Mg2+ [PBS(−)] for 30 min to be dissociated at the single-cell level. PGCs labeled with Venus fluorescence were isolated using a micropipette and transferred into 60% Leibovitz's L-15 medium (Gibco) containing 0.5 mg/mL bovine serum albumin (BSA). They were then cultured at 23°C in the same medium in a petridish (Falcon) or a culture chamber, μ-slide I (ibidi), coated with 1 μg/mL or 0.5 μg/mL fibronectin, respectively (Asahi Glass). For pharmacological analysis, PGCs were incubated in the medium containing 1 μmol/L Cytocharasin D (Cosmobio), 10 μmol/L Blebbistatin (Sigma), 10 μmol/L nocodazol (Sigma), 200 μmol/L NSC23766 (Tocris) or 50 μmol/L Y27632 (Wako). The morphology of PGCs was observed 30 min after adding the chemicals.

Immunological detection of PGCs

Embryos were fixed with 8.8% formaldehyde-0.9% picric acid for 2 h. The fixed embryos were dehydrated through a series of ethanol-butanol solutions, embedded in paraffin, sliced serially 8 μm thick with a microtome, and mounted on glass slides (Matsunami). The rehydrated sections were washed in PBS(−) containing 0.1% Triton X-100 (TPBS) and incubated in 10 mmol/L citric acid buffer (pH 6.0) at 95°C for 20 min. They were then blocked in 10% goat serum in TPBS, and incubated with anti-Xenopus Daz-like protein antibody (Mita & Yamashita 2000). After washing in TPBS, the sections were incubated with Cy3-labeled secondary antibody. After further washing in TPBS, the sections were mounted in 50% glycerol containing Hoechst 33342 (1 μg/mL) and observed under a fluorescence microscope (Olympus BX60). The 3D movies were constructed from images of serial sections using Delta Viewer 2.1 software ( Three embryos were used for each stage, showing similar results.

Rhodamine-phalloidin staining

Isolated PGCs and endodermal cells were cultured on the μ-slide I for about 30 min and fixed in 7.4% formalin in PBS(−) for 30 min. The cells were washed three times in TPBS and incubated with 10 U/mL rhodamine-phalloidin (Molecular Probes) for 30 min in the dark. The cells were washed three times in PBS(−), mounted in 50% glycerol, and observed under a confocal microscope, LSM 510 (Zeiss).

Time-lapse movies and real-time imaging

Living PGCs and endodermal cells were observed with an inverted phase contrast microscope, IMT-2 (Olympus) and BZ-8000 (Keyence). Images were captured at 10 s, 1 min or 2 min intervals for time-lapse movies by using a digital camera (moticam 2000, Motic). The movies were generated by Motion SWF and Swf2Avi software.


Xenopus PGCs transit four migration states in tailbud embryos

The localization of PGCs in Xenopus tailbud embryos has been examined by Kamimura et al. (1976) using morphological characters and by Nishiumi et al. (2005) using a lineage tracer. To better understand the localization of PGCs, embryos at stage 24–41 were sectioned serially and immunostained with the antibody against Xenopus Daz-like protein, which is specific for the PGCs (Mita & Yamashita 2000; Houston & King 2000).

At stage 24, most PGCs made up a small number of clusters deep in the endoderm while a few were separated on their own as previously shown by Kamimura et al. (1976) (Fig. 1a,e; Movie 1). We refer to these cells as PGCs at the clustering stage. Clusters were also observed in stage 18 embryos (date not shown). At stage 28, many individual PGCs were located away from the clusters that still included significant numbers of PGCs (Fig. 1b,f; Movie 2). Some individual PGCs were elongated at this stage. We refer to these cells as PGCs at the dispersing stage. At stage 33/34, the PGCs were mostly scattered in the dorsal half of the endoderm, and rarely found in the ventral half. As they seem to migrate toward the dorsal side at this stage, we refer to them as PGCs at the directionally migrating stage (Fig. 1c,g; Movie 3). At stage 41, more than half the PGCs were located in the most dorsal region of the endoderm, forming clusters (Fig. 1d,h; Movie 4). They were not found in the mesentery at this stage, while previous reports showed that several PGCs were found in the nephric region (Kamimura et al. 1976). We refer to these cells as PGCs at the re-aggregating stage. These observations reconfirmed that the distribution of PGCs in the endoderm changed according to the developmental stage, suggesting a presence of dynamic changes in PGC motility during development.

Figure 1.

Localization of primordial germ cells (PGCs) in tailbud embryos. (a–d) Transverse section of embryos at stage 24 (a), 28 (b), 33/34 (c) and 41 (d). PGCs were labeled with anti-Xenopus Daz-like protein antibody (magenta; a–d). Nuclei were labeled with Hoechst 33342 (green; a′–d′) 3D images of the in vivo localization of PGCs at stage 24 (e), 28 (f), 33/34 (g) and 41 (h) were constructed with Delta Viewer. PGCs, red. Dorsal side up in all photos, with anterior to the left. (e′–h′) Schematic diagrams of localization of PGCs in tailbud embryos. Very small Dazl-positive fragments did not contain nuclei. PGCs, magenta. Endodermal region, yellow. Scale bars: 100 μm.

Xenopus PGCs change their morphology during development

To characterize the PGCs at the four stages, we isolated the PGCs from embryos at stages 18, 24, 28, 33/34, 37/38 and 41. PGCs were labeled by microinjecting Venus-DS mRNA into the fertilized egg (Kataoka et al. 2006). The number of PGCs isolated from a single embryo at each stage was roughly consistent with that in previous reports analyzing serial sections (Kamimura et al. 1976, 1980) (Table S1). The isolated PGCs and neighboring somatic endodermal cells were cultured in a fibronectin-coated chamber with a serum-free medium for 30 min and observed by phase-contrast microscopy. After the short-term observation of these PGCs in vitro, we classified them with morphological criteria into four types of cells, namely: (i) round PGCs with small blebs, (ii) elongated PGCs, (iii) rugged PGCs, and (iv) round PGCs. Interestingly, the ratio of these cell types was dependent upon developmental stages as mentioned below.

All the PGCs isolated from stage 18 embryos and most from stage 24 embryos were classified as the round PGCs with small blebs (Fig. 2a; Table 1). They had several small blebs, which were 12.7 ± 5.2 μm in average length and 10.6 ± 3.0 μm in average diameter. As endodermal cells from embryos at these stages showed the same morphology as the PGCs, we could not distinguish them morphologically (Fig. 2e). The elongated PGCs dominated at stage 28 (64.9%), whereas a certain number of cells were rugged (27.2%) at this stage. The elongated PGCs obviously locomoted in the longitudinal direction in vitro (Fig. 2b; Table 1). A large membrane bleb with clear cytoplasm was observed at the leading front of the elongated PGC. The bleb was 27.8 ± 2.7 μm in average diameter and referred to as a large bleb. A knob-like structure was found at the posterior pole of the elongated PGCs. The rugged PGCs were roughly round but had several large protrusions, one of which was clear and quite similar to the large bleb observed in the elongated PGCs (Fig. 2c). Endodermal cells at stage 28 were round in shape without any obvious protrusions (Fig. 2f). The endodermal cells did not change their morphology from stage 28 to 41. Most of the PGCs isolated from stage 33/34 embryos were rugged (82.0%), while the others were elongated (18.0%). The round PGCs with no protrusions were found in the culture at stage 37/38 and increased at stage 41 (Fig. 2d; Table 1). The cells were almost completely round and indistinguishable in morphology from the endodermal cells at these stages. Thus these four types of cultured PGCs, the round cells with small blebs, the elongated cells, the rugged cells and the round cells dominate successively during the development. This finding strongly suggests that they represent PGCs at the clustering, dispersing, directionally migrating and re-aggregating stages, respectively.

Table 1. Morphological difference of primordial germ cells (PGCs) at each developmental stage
StageNumber of embryos examinedNumber of PGCs examinedCell type (%)
Round PGC with small blebsElongated PGCRugged PGCRound PGC
  1. PGCs were observed 30 min after starting the culture. Each value is the result of six experiments.

1830504100 0 0 0
2431753 91.1 7.8 1.1 0
28391246 8.864.027.2 0
33/34341248 018.082.0 0
37/38341465 0 7.252.840.0
40301369 0 041.158.9
Figure 2.

Morphology of isolated primordial germ cells (PGCs) and somatic endodermal cells. (a–f) PGCs isolated from embryos at stages 24 (a), 28 (b), 33/34 (c) and 41 (d), and endodermal cells isolated from embryos at stages 24 (e) and 28 (f) were cultured in the fibronectin-coated culture chamber. Insets show fluorescence images of the cells. The PGCs were labeled with Venus fluorescence. Arrows indicate the small blebs. Arrowheads indicate the large clear blebs. Images were taken after cells adhered on the substrate. Scale bars: 20 μm.

Xenopus PGCs change their locomotive activity during development

To examine the locomotive properties of the four types of PGCs, we analyzed their behavior using time-lapse video imaging for 3 h. PGCs with small blebs at stage 24 hardly changed their position on the fibronectin substrate, but protruded and retracted the small blebs continuously (Fig. 3a; Movie 5). The same morphological change was also observed on the endodermal cells (Fig. 3e). Elongated PGCs at stage 28 locomoted in the longitudinal direction at an average speed of 704 ± 46 μm/h (Fig. 3b). At the posterior end, a small number of transparent small protrusions were formed and retracted near the knob-like structure (Movie 6). A large clear bleb was also observed on the rugged PGCs, but was not maintained on one side of the cell. It moved around the cell circumference continuously without retracting. Consequently the PGCs seem to form multiple large protrusions in various directions and hardly change their position (Fig. 3c; Movie 7).

Figure 3.

Time-lapse observation of isolated primordial germ cells (PGCs) and somatic cells. (a–f) A PGC isolated from stage 24 (a), 28 (b), 33/34 (c) or 41 (d), or an endodermal cell from stage 24 (e) or 28 (f) was cultured in the fibronectin-coated culture chamber. Arrows indicate small blebs. Arrowheads indicate the large clear blebs. Photos were taken at 10 s intervals after PGCs adhered to the substrate. Scale bars: 20 μm.

The elongated PGCs at stage 28 occasionally changed to the rugged type and stopped locomoting (Fig. 4a). After a certain period, they re-elongated and started locomoting again. On the other hand, the rugged cells, classified at the initial observation, changed into elongated cells and started locomoting after a certain period of cultivation. Therefore, most of the PGCs at stage 28 repeated the alternative phases, namely the locomotive phase with elongated morphology and the pausing phase with rugged morphology (Fig. 4b; Movie 8). The mean locomotive period for each cell ranged from 25 to 65 min (Fig. 4c; Table 2). PGCs isolated from stage 33/34 embryos also alternated between the locomotive and pausing phases. Speed of locomotion was comparable to that at stage 28 (Table 2). However, duration of the locomotive phase was shorter than that of PGCs at stage 28 (Fig. 4b). The mean locomotive period for each cell ranged from 10 to 35 min, overlapping with that at stage 28 (Fig. 4c). Therefore, it seems that there is no discrete population with a long or short duration. Round PGCs appeared at stage 37/38 and increased their population at stage 41. They showed no protrusive or locomotive activities and were indistinguishable in morphology from the endodermal cells at these stages (Fig. 3d,f). In conclusion, the time-lapse analysis revealed that the motility of PGCs changed dramatically during development, and that most PGCs from stage 28 to stage 33/34 have significant locomotive activity on the fibronectin substrate.

Table 2. Characteristics of migrating primordial germ cells (PGCs)
StageNumber of PGCs examinedAverage duration of locomotive phase ± SD (min)Average duration of pausing phase ± SD (min)Average migration speed during locomotive phase (μm/h)a
  1. Time-lapse images of the PGCs were captured at 1 min intervals for 30 min after 1 h of cultivation. Values were calculated using the images.

  2. a

    Average migration speed during only the locomotive phase.

    SD, standard deviation.

285641 ± 1815 ± 6752 ± 44
33/345215 ± 924 ± 12724 ± 43
Figure 4.

Migrating primordial germ cells (PGCs) alternate between locomotive and pausing phases. (a) Time-lapse images of a cultured PGC isolated from a stage 28 embryo. Photos were taken at 4 min intervals. Bar, 50 μm. (b) Change in the locomotion speed of cultured PGCs. Locomotion speed was determined every 2 min. White areas indicate locomotive phase and gray areas indicate pausing phase. The boundary between the locomotive and pausing phases was determined at the time point of half-maximal speed. (c) The graph represents the distribution of the average duration of the locomotive phase of PGCs at stage 28 (gray bars) and 33/34 (white bars).

PGC motility is dependent on actin polymerization and myosin activity

The large bleb observed on the elongated PGCs is thought to be involved in the locomotion in vitro like in many types of migratory cells forming blebs (Fackler and Grosse 2008). Actin polymerization and myosin-based contractions are essential for the bleb to form in various cells (Charras 2008). First, we examined the localization of F-actin in cultured PGCs by rhodamine-phalloidin staining. In elongated PGCs, F-actin was concentrated at the border between the elongated cell body and knob-like region but not found at the leading front (Fig. 5b). On the other hand, it was located at the cell periphery in other types of PGCs (Fig. 5a,c,d). Next, in order to evaluate the role of actin polymerization and myosin activity, we used specific inhibitors (Fig. 5e–h). The elongated PGCs treated with cytochalasin D, an inhibitor of actin polymerization, became round without any protrusions and stopped locomoting (Fig. 5f). Treatment with blebbistatin, an inhibitor of non-muscle myosin II ATPase, resulted in the same morphology as that of cells treated with cytochalasin D (Fig. 5g). On the other hand, microtubule depolymerization induced by nocodazole had no effect on the morphology and motility of the elongated PGCs (Fig. 5h). For rugged PGCs, treatment with these inhibitors resulted in the same morphological change (data not shown). These results demonstrated that both actin polymerization and myosin activity were essential for the bleb formation and cell elongation, and were required for the locomotion of Xenopus PGCs.

Figure 5.

Localization of F-actin and roles of cytoskeletons in migrating primordial germ cells (PGCs). (a–d) Localization of F-actin in isolated PGCs. A representative PGC isolated from an embryo at each stage was stained with rhodamine-phalloidin. Arrows in (a) indicate the small blebs. Arrows in (b) indicate the rear region. Arrowheads in (b) indicate the front of an elongated PGC. Arrows in (c) indicate the large bleb. (e–h) Elongated PGCs at stage 28 were treated with indicated inhibitors. Photos were taken 30 min after adding the chemicals. Scale bars: 20 μm.

PGC migration is dependent on ROCK/RhoA signaling

Rho family small GTPases, including RhoA, Rac1 and Cdc42, have been shown to regulate actin polymerization in various types of migrating cells (Ridley 2001). RhoA and its downstream effector, Rho-associated protein kinase (ROCK), are both required for the formation of stress fibers and focal adhesions (Ridley 2001). In addition to the above roles, RhoA/ROCK signaling regulates bleb formation by phosphorylating myosin light chain (MLC) (Fackler & Grosse 2008). Rac1 and Cdc42 regulate the formation of lamellipodia and filopodia, respectively (Nobes & Hall 1995). To evaluate the roles of the small GTPases in PGC migration, we injected mRNA for their dominant-negative forms flanked by the DEADSouth 3'UTR which restricted the gene expression in PGCs (Kataoka et al. 2006) and avoided global effects of the dominant-negative mutants including the gastrulation defect. When only Venus-DS-injected embryos were observed externally at stage 41, the PGCs were found to be aligned around the dorsal body wall region, as described previously (Kataoka et al. 2006) (Fig. 6a,i). Overexpression of the dominant-negative RhoA (RhoA-DN) completely blocked the normal PGC localization (Fig. 6b,i). Dominant-negative Rac1 (Rac1-DN) reduced the number of embryos with correctly migrated PGCs in a dose-dependent manner (Fig. 6c,i). Dominant-negative Cdc42 (Cdc42-DN) did not affect the PGC localization (Fig. 6d,i). These results suggested that RhoA is essential for the proper migration of Xenopus PGCs in vivo.

Figure 6.

Effects of inhibiting Rho family GTPase signals on primordial germ cell (PGC) migration. (a–d) Lateral views of stage 41 embryos injected with Venus-DS mRNA (460 pg) (a), or co-injected with Venus-DS mRNA (460 pg) and mRNA for RhoA-DN (460 pg) (b), Rac1-DN (460 pg) (c) or Cdc42-DN (2.3 ng) (d). Dorsal side up and anterior to the left. (e, f) Isolated PGCs from stage 28 embryos injected with RhoA-DN (460 pg) or Rac1-DN (460 pg) were cultured on fibronectin-coated petridishes. (g, h) Elongated PGCs isolated from stage 28 embyos were incubated with Y27632 (ROCK inhibitor) or NSC27632 (Rac1 inhibitor). Photos were taken 30 min after adding the chemicals. Arrowheads indicate the PGCs in the normal region. Arrows indicate the ectopic PGCs. Scale bars: 1 mm in d; 20 μm in (e–h). (i) Effects of DN-small GTPases on PGC localization in vivo. Venus-DS mRNA (460 pg) was injected into all the embryos for PGC labeling, and additionally indicated mRNAs were injected. PGC localization was observed externally from both sides at stage 41. Graph shows the percentages of embryos with indicated PGCs. Values are includes results of three experiments. (j, k) Effects of DN-small GTPases on elongation (j) and locomotion (k) of isolated PGCs. Time-lapse images of the stage 28 PGCs were captured at 1 min intervals for 30 min after 1 h of cultivation. If a PGC showed an elongated shape during the culture period, it was counted as an elongated PGC. If a PGC moved more than 100 μm for 30 min, it was counted as a locomoting PGC. Values are results of three experiments. C; control injected with only Venus-DS mRNA. *P < 0.05, **P < 0.001.

To gain further insight into their roles in the migrating Xenopus PGCs, we isolated PGCs from stage 28 embryos that had been injected with Rac1-DN, RhoA-DN or Cdc42-DN and cultured them to analyze their morphology and locomotive properties. The PGCs from RhoA-DN-injected embryos in which PGC locomotion was completely blocked, showed a round morphology without any protrusions (Fig. 6e,j). They were morphologically very similar to PGCs treated with cytocharasin D or blebbistatin. The percentage of locomoting PGCs was reduced by Rac1-DN in a dose-dependent manner, although that of elongated PGCs was not affected (Fig. 6f,j,k). It appeared that a certain number of elongated PGCs could not detach from the substrate and could not change their position. On the other hand, Cdc42-DN did not affect PGC morphology or locomotion (Fig. 6j,k). No dominant negative form had any influence on the number or size of PGCs (data not shown), suggesting that the proliferation, cell death or differentiation of PGCs was not affected significantly. To obtain additional evidence on the roles of the RhoA signal and Rac1, we performed a pharmacological analysis. Treatment with Y27632, a ROCK inhibitor, abolished the locomotion of PGCs, corresponding to the effect of RhoA-DN (Fig. 6g). On the other hand, NSC23766, a Rac1 inhibitor, did not affect the morphology or locomotive properties of the PGCs (Fig. 6h). From these results, we concluded that ROCK/RhoA signaling regulates the locomotion of the Xenopus PGCs.


From histochemical and ultrastructural observations, it has been suggested that PGCs start to migrate at the tailbud stage in Xenopus (Nishiumi et al. 2005; Kamimura et al. 1980). However, it had been unclear whether the PGC itself acquires its motility or whether the environment becomes permissive for the migration at that time. Here we have successfully cultivated living PGCs isolated from Xenopus embryos at various stages of development. We also showed that the morphology and locomotive activity of PGCs drastically changed, depending upon the developmental stage. The characteristics of PGCs in vitro appeared to correspond to the state of the PGCs in vivo as discussed below.

Correlation between properties in vitro and state of PGCs in vivo

Primordial germ cells from stage 18 and 24 embryos are indistinguishable in morphological criteria from somatic endodermal cells at the same stage. The PGCs showed no locomotion on the fibronectin substrate, suggesting that they have little migratory activity at this stage. Our finding is consistent with the previous suggestion that PGCs would migrate passively with surrounding endodermal cells until stage 23/24 (Nishiumi et al. 2005).

Most PGCs from stage 28 or 33/34 embryos alternated between the locomotive phase with the elongated morphology and the pausing phase with the rugged morphology. PGCs at stage 28 showed the elongated morphology for most of the culture period, and rapidly moved on the fibronectin substrate. Brustis et al. (1984) reported that a subpopulation of cells in a culture of endodermal cell mass from stage 29 embryos had an elongated morphology, and suggested them to be PGCs. Our observation supports their suggestion. Our localization study suggests that PGCs in stage 28 embryos are separating from the cluster. The robust locomotive activity of the elongated PGCs in the culture is thought to be a major force leading to the separation in vivo.

Primordial germ cells in the pausing phase exhibited the rugged morphology. Tarbashevich et al. (2011) reported that PGCs from stage 32 embryos were round with a bleb circulating in the periphery although the morphology was clearly different from that of the rugged PGCs. The circular movement of the bleb is also observed by us not only on PGCs but also on certain numbers of somatic cells in the early phase of the cultivation when the cells do not adhere to the substrate yet (data not shown). Therefore the circular movement is not a PGC-specific character at least in our experiments. The difference is attributed to the difference in the culture conditions and/or preparation method in that they used trypsin and ethylenediaminetetraacetic acid (EDTA), whereas we did not.

The round PGCs without any detectable motility were dominant in the culture prepared from the embryos at stage 41. This suggests that most of the PGCs at the earlier stage had lost their locomotive activity by that time. A considerable number of PGCs are located in the most dorsal region of the endoderm to form clusters. In later stages, for example, at stage 43, most PGCs are located in the dorsal mesentery and surrounded by the mesenchymal cells (Heasman & Wylie 1978). Although the mechanism underlying the translocation from the dorsal endoderm to the mesentery is unclear, Whitington & Dixon (1975) suggested that the PGCs would be carried passively from the endoderm by the mesentery cells. This suggestion is consistent with our finding that most of the PGCs located at the dorsal endoderm are immotile. PGCs on the mesentery cells at stage 42/43 showed migrating behavior both in vivo and in vitro (Wylie & Heasman 1976; Heasman et al. 1981). The immotile PGCs may reactivate their migration in the mesentery. Alternatively, PGCs with the same potential may exhibit quite different migration behaviors depending on the environment in which they exist.

Directional migration of Xenopus PGCs

Primordial germ cells at stage 28 appeared to be separating from the clusters with little directionality. Thus, the mechanisms giving rise to the directional migration of PGCs, such as chemotaxis, may not have significant roles at this stage. At stage 33/34, most PGCs were scattered throughout the dorsal half of the endoderm, and PGCs were rarely seen in the ventral half, suggesting that they migrate directionally within the endodermal cell mass toward the dorsal part. Our in vitro observations showed that most PGCs from stage 28 embryos exhibited long locomotive and short pausing phases, while the majority of stage 33/34 PGCs have short locomotive and long pausing phases. This result supports the idea that the pausing phase along with the rugged behavior is involved in the mechanism underlying the directional migration. The alternation of locomotive and pausing phases at this stage may correspond to repeats of the run and the tumbling phases in zebrafish PGCs (Reichman-Fried et al. 2004). It remains to be determined whether the alternate repeat of the two phases plays a significant role in the directional migration.

Mechanisms of PGC locomotion

Cellular blebbing is associated with the migration behavior of several species, including Amoeba and Dictyostelium. As the Amoeba proteus migrates extending a large bleb at its front, it was suggested that the driving force of the migration is generated mainly by the actomyosin contraction at the rear (Yanai et al. 1996). This is consistent with our observation that F-actin is concentrated at the rear of the elongated locomoting PGC. Expansion of the bleb in a migrating Dictyostelium cell is promoted by hydrostatic pressure generated by cortical contraction involving the activity of myosin II (Yoshida & Inouye 2001). Non-muscle type myosin II has been shown to play an essential role in various migrating cells with cellular blebs, including invasive tumor cells (Poincloux et al. 2011) and zebrafish PGCs (Blaser et al. 2006; Goudarzi et al. 2012). It is also critical to PGC migration in Xenopus because both the blebbing and the in vitro locomotion are completely blocked by treatment with the myosin II inhibitor, blebbistatin. The migrating PGCs of Xenopus and zebrafish share several characteristics including the formation of blebs, dependency on myosin II and RhoA, and cytoplasmic flow into the blebs, suggesting conserved mechanisms underlying PGC migration between the two species. One distinct difference is that the elongated Xenopus PGCs at stage 28 continue to protrude the bleb in one direction for a considerable period, whereas in zebrafish PGCs, the site of the bleb changes frequently.

In addition to the essential role of RhoA, our analysis using dominant-negative mutants showed a moderate inhibition of the PGC migration in vivo by Rac1-DN. Observations in vitro showed that Rac1 inhibition does not affect the elongation or blebbing, but appears to partially suppress the detachment of PGCs from the substrate in the locomotive process. Consistent with our observation, it has been reported that Rac1 plays a role in the regulation of cell adhesive behavior on fibronectin (Hens et al. 2002). As NSC23766 had no effect on PGC locomotion in vitro, the effect of Rac1-DN should be verified by further analysis to reveal the precise role of Rac1 in the PGC migration.

Does the cultured PGC reflect cellular behavior in vivo?

Our in vitro study clearly shows that the motility of PGCs changes dramatically during development and suggests a correlation between the in vitro behavior of the PGC and its status in vivo. The environment surrounding cultured PGCs, however, is quite different from that in the embryo; PGCs adhere to the fibronectin substrate without contacting any other cell in vitro, whereas they are surrounded by somatic cells and/or other PGCs in the embryo. A substrate, such as fibronectin or collagen, is required for the adhesion and locomotion of the isolated PGCs in culture (data not shown). An immunohistochemical analysis showed that the PGCs synthesize fibronectin from stage 26 to 40 in the Xenopus embryo (Nishiumi et al. 2005). Therefore it is possible that fibronectin has a significant role in PGC migration both in vivo and in vitro. On the other hand, cell-to-cell adhesion mediated by cell adhesion molecules, such as cadherins, may be involved in Xenopus PGC migration, as it has a significant role in the PGC migration within the gut of the mouse embryo (Bendel-Stenzel et al. 2000).

The locomoting PGC in vitro does not protrude a filopodia or a lamellipodia, which is the characteristic structure found on many migrating cells and involved in the migration of fibroblastic cells. We cannot exclude the possibility that the PGC forms cytoplasmic protrusions under normal conditions but not in culture. On the other hand, the formation of a lamellipodium and/or filopodium appears to be unnecessary in many migrating processes dependent on myosin II and RhoA activities (Charras & Paluch 2008). Despite difficulties due to the large number of yolk platelets and melanosomes included in embryonic Xenopus cells, a future study using live imaging of migrating PGCs in whole embryos or in a slice culture should clarify the differences between PGCs in culture and those in embryos.


This work was supported by JSPS KAKENHI Grant Number 23570269, 24570239. We are grateful to Dr Yamashita (Graduate School of Science, Hokkaido University) for generously providing the anti-Xdazl monoclonal antibody. We also thank the members of our laboratory for discussions and technical support.